Development of sustainable groundwater management methodologies to control saltwater intrusion into coastal aquifers with application to a tropical Pacific island country

Lal, Alvin Abinesh (2019) Development of sustainable groundwater management methodologies to control saltwater intrusion into coastal aquifers with application to a tropical Pacific island country. PhD thesis, James Cook University.

PDF (Thesis)
Download (4MB) | Preview
View at Publisher Website:


Saltwater intrusion due to the over-exploitation of groundwater in coastal aquifers is a critical challenge facing groundwater-dependent coastal communities throughout the world. Sustainable management of coastal aquifers for maintaining abstracted groundwater quality within permissible salinity limits is regarded as an important groundwater management problem necessitating urgent reliable and optimal management methodologies. This study focuses on the development and evaluation of groundwater salinity prediction tools, coastal aquifer multi-objective management strategies, and adaptive management strategies using new prediction models, coupled simulation-optimization (S/O) models, and monitoring network design, respectively.

Predicting the extent of saltwater intrusion into coastal aquifers in response to existing and changing pumping patterns is a prerequisite of any groundwater management framework. This study investigates the feasibility of using support vector machine regression (SVMR), an innovative artificial intelligence-based machine learning algorithm, to predict salinity at monitoring wells in an illustrative aquifer under variable groundwater pumping conditions. For evaluation purposes, the prediction results of SVMR are compared with well-established genetic programming (GP) based surrogate models. The prediction capabilities of the two learning machines are evaluated using several measures to ensure their practicality and generalisation ability. Also, a sensitivity analysis methodology is proposed for assessing the impact of pumping rates on salt concentrations at monitoring locations. The performance evaluations suggest that the predictive capability of SVMR is superior to that of GP models. The sensitivity analysis identifies a subset of the most influential pumping rates, which is used to construct new SVMR surrogate models with improved predictive capabilities. The improved predictive capability and generalisation ability of SVMR models, together with the ability to improve the accuracy of prediction by refining the dataset used for training, make the use of SVMR models more attractive.

Coupled S/O models are efficient tools that are used for designing multi-objective coastal aquifer management strategies. This study applies a regional-scale coupled S/O methodology with a Pareto front clustering technique to prescribe optimal groundwater withdrawal patterns from the Bonriki aquifer in the Pacific Island of Kiribati. A numerical simulation model is developed, calibrated and validated using field data from the Bonriki aquifer. For computational feasibility, SVMR surrogate models are trained and tested utilizing input-output datasets generated using the flow and transport numerical simulation model. The developed surrogate models were externally coupled with a multi-objective genetic algorithm optimization (MOGA) model, as a substitute for the numerical model. The study area consisted of freshwater pumping wells for extracting groundwater. Pumping from barrier wells installed along the coastlines is also considered as a management option to hydraulically control saltwater intrusion. The objective of the multi-objective management model was to maximise pumping from production wells and minimize pumping from barrier wells (which provide a hydraulic barrier) to ensure that the water quality at different monitoring locations remains within pre-specified limits. The executed multi-objective coupled S/O model generated 700 Pareto-optimal solutions. Analysing a large set of Pareto-optimal solution is a challenging task for the decision-makers. Hence, the k-means clustering technique was utilized to reduce the large Pareto-optimal solution set and help solve the large-scale saltwater intrusion problem in the Bonriki aquifer.

The S/O-based management models have delivered optimal saltwater intrusion management strategies. However, at times, uncertainties in the numerical simulation model due to uncertain aquifer parameters are not incorporated into the management models. The present study explicitly incorporates aquifer parameter uncertainty into a multi-objective management model for the optimal design of groundwater pumping strategies from the unconfined Bonriki aquifer. To achieve computational efficiency and feasibility of the management model, the calibrated numerical simulation model in the S/O model was is replaced with ensembles of SVMR surrogate models. Each SVMR standalone surrogate model in the ensemble is constructed using datasets from different numerical simulation models with different hydraulic conductivity and porosity values. These ensemble SVMR models were coupled to the MOGA model to solve the Bonriki aquifer management problem for ensuring sustainable withdrawal rates that maintain specified salinity limits. The executed optimization model presented a Pareto-front with 600 non-dominated optimal trade-off pumping solutions. The reliability of the management model, established after validation of the optimal solution results, suggests that the implemented constraints of the optimization problem were satisfied; i.e., the salinities at monitoring locations remained within the pre-specified limits.

The correct implementation of a prescribed optimal management strategy based on the coupled S/O model is always a concern for decision-makers. The management strategy actually implemented in the field sometimes deviates from the recommended optimal strategy, resulting in field-level deviations. Monitoring such field-level deviations during actual implementation of the recommended optimal management strategy and sequentially updating the strategy using feedback information is an important step towards adaptive management of coastal groundwater resources. In this study, a three-phase adaptive management framework for a coastal aquifer subjected to saltwater intrusion is applied and evaluated for a regional-scale coastal aquifer study area. The methodology adopted includes three sequential components. First, an optimal management strategy (consisting of groundwater extraction from production and barrier wells) is derived and implemented for the optimal management of the aquifer. The implemented management strategy is obtained by solving a homogeneous ensemble-based coupled S/O model. Second, a regional-scale optimal monitoring network is designed for the aquifer system, which considers possible user noncompliance of a recommended management strategy and uncertainty in aquifer parameter estimates. A new monitoring network design is formulated to ensure that candidate monitoring wells are placed at high risk (highly contaminated) locations. In addition, a k-means clustering methodology is utilized to select candidate monitoring wells in areas representative of the entire model domain. Finally, feedback information in the form of salinity measurements at monitoring wells is used to sequentially modify pumping strategies for future time periods in the management horizon. The developed adaptive management framework is evaluated by applying it to the Bonriki aquifer system. Overall, the results of this study suggest that the implemented adaptive management strategy has the potential to address practical implementation issues arising due to user noncompliance, as well as deviations between predicted and actual consequences of implementing a management strategy, and uncertainty in aquifer parameters.

The use of ensemble prediction models is known to be more accurate standalone prediction models. The present study develops and utilises homogeneous and heterogeneous ensemble models based on several standalone evolutionary algorithms, including artificial neural networks (ANN), GP, SVMR and Gaussian process regression (GPR). These models are used to predict groundwater salinity in the Bonriki aquifer. Standalone and ensemble prediction models are trained and validated using identical pumping and salinity concentration datasets generated by solving numerical 3D transient density-dependent coastal aquifer flow and transport numerical simulation models. After validation, the ensemble models are used to predict salinity concentration at selected monitoring wells in the modelled aquifer under variable groundwater pumping conditions. The predictive capabilities of the developed ensemble models are quantified using standard statistical procedures. The performance evaluation results suggest that the predictive capabilities of the standalone prediction models (ANN, GP, SVMR and GPR) are comparable to those of the groundwater variable-density flow and salt transport numerical simulation model. However, GPR standalone models had better predictive capabilities than the other standalone models. Also, SVMR and GPR standalone models were more efficient (in terms of computational training time) than other standalone models. In terms of ensemble models, the performance of the homogeneous GPR ensemble model was found to be superior to that of the other homogeneous and heterogeneous ensemble models.

Employing data-driven predictive models as replacements for complex groundwater flow and transport models enables the prediction of future scenarios and also helps save computational time, effort and requirements when developing optimal coastal aquifer management strategies based on coupled S/O models. In this study, a new data-driven model, namely Group method for data handling (GMDH) approach is developed and utilized to predict salinity concentration in a coastal aquifer and, simultaneously, determine the most influential input predictor variables (pumping rates) that had the most impact onto the outcomes (salinity at monitoring locations). To confirm the importance of variables, three tests are conducted, in which new GMDH models are constructed using subsets of the original datasets. In TEST 1, new GMDH models are constructed using a set of most influential variables only. In TEST 2, a subset of 20 variables (10 most and 10 least influential variables) are used to develop new GMDH models. In TEST 3, a subset of the least influential variables is used to develop GMDH models. A performance evaluation demonstrates that the GMDH models developed using the entire dataset have reasonable predictive accuracy and efficiency. A comparison of the performance evaluations of the three tests highlights the importance of appropriately selecting input pumping rates when developing predictive models. These results suggest that incorporating the least influential variables decreases model accuracy; thus, only considering the most influential variables in salinity prediction models is beneficial and appropriate.

This study also investigated the efficiency and viability of using artificial freshwater recharge (AFR) to increase fresh groundwater pumping rates from production wells. First, the effect of AFR on the inland encroachment of saline water is quantified for existing scenarios. Specifically, groundwater head and salinity differences at monitoring locations before and after artificial recharge are presented. Second, a multi-objective management model incorporating groundwater pumping and AFR is implemented to control groundwater salinization in an illustrative coastal aquifer system. A coupled SVMR-MOGA model is developed for prescribing optimal management strategies that incorporate AFR and groundwater pumping wells. The Pareto-optimal front obtained from the SVMR-MOGA optimization model presents a set of optimal solutions for the sustainable management of the coastal aquifer. The pumping strategies obtained as Pareto-optimal solutions with and without freshwater recharge shows that saltwater intrusion is sensitive to AFR. Also, the hydraulic head lenses created by AFR can be used as one practical option to control saltwater intrusion. The developed 3D saltwater intrusion model, the predictive capabilities of the developed SVMR models, and the feasibility of using the proposed coupled multi-objective SVMR-MOGA optimization model make the proposed methodology potentially suitable for solving large-scale regional saltwater intrusion management problems.

Overall, the development and evaluation of various groundwater numerical simulation models, predictive models, multi-objective management strategies and adaptive methodologies will provide decision-makers with tools for the sustainable management of coastal aquifers. It is envisioned that the outcomes of this research will provide useful information to groundwater managers and stakeholders, and offer potential resolutions to policy-makers regarding the sustainable management of groundwater resources. The real-life case study of the Bonriki aquifer presented in this study provides the scientific community with a broader understanding of groundwater resource issues in coastal aquifers and establishes the practical utility of the developed management strategies.

Item ID: 64261
Item Type: Thesis (PhD)
Keywords: adaptive coastal aquifer management, aquifer parameter uncertainty, artificial freshwater recharge, Bonriki aquifer, coastal aquifer management strategies, coastal aquifers, compliance monitoring, ensemble prediction models, generalisation ability, genetic programming, GMDH prediction models, groundwater salinity, groundwater, heterogeneous ensemble model, homogeneous ensemble model, k-means clustering, monitoring network design, multi-objective management model, Pacific island, pareto-optimal solutions, saltwater intrusion, sensitivity analysis, simulation-optimization, support vector regression, surrogate model, variable selection
Related URLs:
Copyright Information: Copyright © 2019 Alvin Abinesh Lal.
Additional Information:

For this thesis, Alvin Lal received the Dean's Award for Excellence 2020, which recognises excellence in Higher Degree by Research and recipients of this award are commended by independent expert examiners as having made a substantial contribution to their field of research.

Seven publications arising from this thesis are stored in ResearchOnline@JCU, at the time of processing. Please see the Related URLs. The publications are:

Chapter 3: Lal, Alvin, and Datta, Bithin (2018) Development and implementation of support vector machine regression surrogate models for predicting groundwater pumping-induced saltwater intrusion into coastal aquifers. Water Resources Management, 32 (7). pp. 2405-2419.

Chapter 4: Lal, Alvin, and Datta, Bithin (2019) Optimal groundwater-use strategy for saltwater intrusion management in a Pacific Island country. Journal of Water Resources Planning and Management, 145 (9). 04019032.

Chapter 5: Lal, Alvin, and Datta, Bithin (2019) Multi-objective groundwater management strategy under uncertainties for sustainable control of saltwater intrusion: solution for an island country in the South Pacific. Journal of Environmental Management, 234. pp. 115-130.

Chapter 6: Lal, Alvin, and Datta, Bithin (2019) Application of monitoring network design and feedback information for adaptive management of coastal groundwater resources. International Journal of Environmental Research and Public Health, 16 (22). 4365.

Chapter 7: Lal, Alvin, and Datta, Bithin (2020) Performance evaluation of homogeneous and heterogeneous ensemble models for groundwater salinity predictions: a regional-scale comparison study. Water, Air and Soil Pollution, 231. 320.

Chapter 8: Lal, Alvin, and Datta, Bithin (2020) Application of the group method of data handling and variable importance analysis for prediction and modelling of saltwater intrusion processes in coastal aquifers. Neural Computing and Applications. (In Press)

Chapter 9: Lal, Alvin, and Datta, Bithin (2017) Modelling saltwater intrusion processes and development of a multi-objective strategy for management of coastal aquifers utilizing planned artificial freshwater recharge. Modeling Earth Systems and Environment, 4 (1). pp. 111-126.

Date Deposited: 03 Sep 2020 23:28
FoR Codes: 09 ENGINEERING > 0905 Civil Engineering > 090509 Water Resources Engineering @ 100%
SEO Codes: 96 ENVIRONMENT > 9609 Land and Water Management > 960903 Coastal and Estuarine Water Management @ 100%
Downloads: Total: 15
Last 12 Months: 15
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page